Applying a passive network reconstruction technique to Twitter data in order to identify trend setters

In this work we apply a systems-theoretic approach to identifying trend setters on Twitter. A network reconstruction algorithm was applied to Twitter data to determine causal relationships among topics discussed by popular Twitter users. Causal relationships in this context means that the topics tweeted by a single user influences the topics tweeted by another user, regardless of sentiment. A user that causally influences other users, without themselves being strongly influenced is identified as a trendsetter. This work seeks to identify potential trendsetters among popular Twitter users and demonstrating that causal influence does not always directly correlate with a user's popularity in terms of followers-demonstrating that popularity alone may not be sufficient for identifying trendsetters on Twitter.